Introduction to Machine Learning for Causal Inference

Introduction to Machine Learning for Causal Inference

A training course organised by the HESG Early Career Researcher Subcommittee

Date: Wednesday 22nd June 2022, 09:30-12:30

Location: HESG 2022 summer meeting, Sheffield


This HESG ECR course has been designed to give health economists an overview and hands-on experience with machine learning (ML) methods. It focuses on recent innovations that enable the use of ML for causal inference, such as ‘causal ML’. Using a mixture of lectures and practical exercises, the short course aims to achieve the following learning objectives:

  • Understand the benefits and limitations of using ML for predictive purposes versus causal inference
  • Understand how predictive learning is used, and for what purposes
  • Gain familiarity with broad ML concepts such as cross-validation and cross fitting
  • Learn how to use causal machine learning to estimate average treatment effects (e.g. using double/debiased ML)
  • Understand the benefits and limitations of using causal ML to estimate heterogeneous treatment effects (e.g. using Causal Forests)

The practical parts of the course will be carried out on R using software packages available on CRAN. The course will be taught by Dr Noemi Kreif (Senior Research Fellow, University of York) and Dr Julia Hatamyar (Research Fellow, University of York). Noemi and Julia have long-standing research experience and interest in using these methods for health policy evaluation.

Practical details

  • The course will take place on 22nd June 2022, as part of the HESG 2022 summer meeting in Sheffield.
  • Please note that the registration for this course is allowed only for ECRs that are already registered for the HESG meeting in Sheffield.
  • The session will last three hours, from 9:30 a.m. to 12:30 p.m. UK time.
  • The course will include lectures and exercises.
  • The exercises will be illustrated by the lecturers using R. While not a formal prerequisite for this course, familiarity with R will enable attendees to gain more from the course by following the demonstrations on their own computers.
  • Attendees are expected to have a laptop with R already installed. Detailed instructions on required R software and package installation, and data to download will be provided in advance. However, support with the installation will not be provided during the course.
  • The course is free thanks to the support of the Health Foundation.
  • The course will be face-to-face only.
  • Support with the practical exercises will be provided by the lecturers if possible.
  • 30 places are available on a first-come-first-served basis.
  • We expect high demand for this face-to-face course, so please ensure that you can attend when you book. We will operate a waiting list.

Early Career Researcher (ECR) definition

For the purposes of this workshop, we define ECRs as follows:

You are an Early Career Researcher (ECR) if you are either (a) currently enrolled in a research degree (e.g. PhD programme) or (b) employed and within 7 years of your highest degree (e.g. Masters or PhD). If you don’t meet these criteria (e.g. due to career breaks), but identify as an ECR, we can take into account any exceptional circumstances.

All participants must also be current members of HESG.